A system uses facial recognition to exclude persons of interest (e.g., “undesirables”) from events and/or venues. Such a system can include a combination of cameras, edge processing devices, and servers that are in communications coupling (e.g., via a network) and that are on premise and/or in the cloud to recognize such persons of interest (POI) and interdict and prevent such POI from entering, traversing, and/or attending (collectively, “entering”) such venues or events.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
2. The system of claim 1, wherein each edge computing device executes multiple software modules that together perform facial detection.
This invention relates to edge computing systems designed for efficient facial detection. The system addresses the challenge of performing real-time facial detection in distributed environments where processing must occur close to data sources to reduce latency and bandwidth usage. Traditional cloud-based facial detection systems often suffer from delays and high data transmission costs, making them unsuitable for applications requiring immediate processing, such as surveillance or augmented reality. The system includes multiple edge computing devices, each executing multiple software modules that collectively perform facial detection. These modules may include preprocessing, feature extraction, and classification components, which work together to identify and analyze faces in images or video streams. By distributing the workload across edge devices, the system ensures low-latency processing and reduces reliance on centralized cloud servers. The modular design allows for flexibility in deployment, enabling customization based on specific use cases or hardware constraints. Additionally, the system may incorporate techniques to optimize resource usage, such as load balancing or adaptive processing, to maintain performance under varying conditions. This approach enhances scalability and reliability, making it suitable for large-scale deployments in smart cities, industrial automation, or consumer electronics.
3. The system of claim 2, wherein the multiple software modules include a face detection module that detects facial images in an image frame and that assigns each face a unique ID.
4. The system of claim 3, wherein the multiple software modules a face tracking module that tracks each face detected by the face detection module from frame to frame in the video stream and that recognizes that the face is either the same one or new entrant into the frame.
This invention relates to a video processing system designed to analyze and track faces in a video stream. The system addresses the challenge of accurately identifying and following individual faces across multiple frames, distinguishing between recurring faces and new entrants. The core functionality involves a face detection module that scans each frame to locate faces, followed by a face tracking module that monitors these faces over time. The face tracking module determines whether a detected face is the same individual appearing in subsequent frames or a new person entering the scene. This capability is essential for applications such as surveillance, user interaction in augmented reality, or behavioral analysis, where consistent face identification is critical. The system may also include additional modules for processing the tracked faces, such as facial recognition or expression analysis, to enhance its utility in various domains. By maintaining continuity of face identity across frames, the system improves the reliability of face-related data extraction in dynamic video environments.
5. The system claim 2, wherein each edge detection device normalizes facial images to a preselected size.
A system for facial recognition or analysis processes facial images using multiple edge detection devices. Each device normalizes the facial images to a preselected size before performing edge detection. Normalization ensures consistent image dimensions, improving accuracy and efficiency in subsequent edge detection. The edge detection devices analyze the normalized images to identify key facial features, such as contours, edges, and boundaries. These features are then used for tasks like facial recognition, expression analysis, or identity verification. The system may include additional components, such as image capture devices or processing units, to enhance performance. By standardizing image size, the system reduces variability in feature detection, leading to more reliable results across different input images. This approach is particularly useful in applications requiring high precision, such as security systems, biometric authentication, or medical imaging. The normalization step ensures that variations in image resolution or scale do not affect the accuracy of edge detection, making the system robust for diverse real-world scenarios.
6. The system of claim 2, wherein the multiple software modules include a face recognizer module to converting each face in the stream of images to a vector and to find a degree of similarity between each such vector and the downloaded embeddings.
The system is designed for facial recognition in video streams, addressing the challenge of accurately identifying individuals in real-time surveillance or security applications. The system processes a continuous stream of images, extracting and analyzing facial features to match against a database of pre-stored facial embeddings. A key component is a face recognizer module that converts each detected face in the image stream into a numerical vector representation. This module then compares these vectors against downloaded embeddings—precomputed facial signatures—to determine the degree of similarity between the detected faces and known identities. The system likely integrates with other modules, such as a face detector to locate faces in the images and a data interface to manage the embedding database. The face recognizer module's ability to quantify similarity enables efficient identification or verification of individuals, supporting applications like access control, surveillance, or personalization. The system may also include preprocessing steps to enhance image quality or handle variations in lighting, pose, or occlusion, ensuring robust performance in real-world scenarios.
7. The system of claim 2, wherein each edge detection device initializes a module to generate a match signal for facial images for which the degree of similarity is above that threshold.
9. An edge computing-based method according to claim 8, comprising transmitting to one or more handheld devices a facial image for which a match signal was generated.
10. An edge computing-based method of claim 9, comprising transmitting to the one or more handheld devices the degree of similarity on which the match signal was based.
11. The edge computing-based method of claim 8, wherein the downloading step includes downloading with the embeddings an indication of an algorithm by which they were generated.
12. The edge computing-based method of claim 11, wherein subsequent to a change in the algorithm the changed algorithm is downloaded to the edge computing device before the embeddings are downloaded.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
March 9, 2021
November 8, 2022
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.